Model Based Mediation With Domain Maps

___________________________• Xiaosen Li (xsli2@uic.edu)

• Guanrao Chen (gchen@cs.uic.edu)

• William Sunna (wsunna1@uic.edu)

• Instructor (Prof. Isabel Cruz)

• The University of Illinois at Chicago

Outline• Introduction• XML-based Mediation• Model-based Mediation• Model-based Mediation with Domain Maps• Application in Bioinformatics• ISIS• Comparison of ISIS and Model-based

Mediation

Different Schemes

Federated Databases XML-Based Mediation Model-Based Mediation

One-World One-/Multiple-Worlds Complex Multiple-Worlds

Our Goal• Given different data sources:

• And we have different queries: (Q1,Q2,……..Qk) over (S1,S2…..Sk)

• Find answers to these questions: (A1,A2,……..Ak)

S1 S2 Sn…….

Introduction

Model-Based Mediation:

Integration of different data sources to retrieve information

that cannot be retrieved using one source.

Domain Map

Domain Maps (Ontology's): Glue Knowledge Sources

“One Simple World” example

• Given: car Dealer A, Car Dealer B

• Find cars from Dealer A and Dealer B, Join on Make. Group by Manufacturing year, and Price.

• Solution: we can use XML-Based Mediation to find the answer.

XML-Mediator (Abstract)

S1

Wrapper Wrapper

XML VIEW XML VIEW

MEDIATOR IntegratedXML View

IVD(S1,S2)

Query(s1,s2)

USER

…... SnS2You can add multiple sources

XML QUERIES/RESULTS

CAR DEALER A

CAR DEALER B

Integrated View Definition for the Car exampleXMAS

XML Matching And Structuring language

CONSTRUCT <Cars>

<car>

$m1

$p

<make>

$ma { $ma }

</make>

</car> { $m1, $p }

</cars>

WHERE <cars.car>$m1 : <Manu_date />$p : <Price/> </> IN

WRAP(“Dealer_A”)

AND< Manu_dates. Manu_date >$m2 : <Manu_date />

<make> $ma : <make/> </> </> IN

WRAP(“Dealer_B”)AND value( $m1 ) = value( $m2 )

XMAS QUERY PROCESSING

Translator COMPOSITIONRewriter,Optimizer

XMAS QUERY

XMAS VIEWDEFINITION

PLAN EXECUTION

XML-Based Mediation:

– XML Models– XML Elements– Structural Constraints:

• DTD (Parent, Child, Sibling)

– No classes relationships (is-a, has-a)– No logical Domain constraints

Complex Multiple-WorldsNavigating the multiresolution data using

knowledge-based mediation with domain maps

___________________________

Genes Proteins Cells Tissues Organism

s

Different Species

Different Techniques

Different Disciplines

Complex Multiple-Worlds

Strategies

• Take all the huge different databases and put them into an even larger database (warehouse)

• Or develop a system to talk to different databases and correlates the results

Database Database Database

What is the cerebellar distribution of rat proteins with more than 80% homolgy with human NCS-1?How about other rodents?

Protein localization Morphology Neurotransmision

Database•••••

CaBP

SYSTEM THAT CAN PROCESS THE QUERYFROM MULTIPLE COMPLEX WORLD

DATABASES

Query/Result

Model-based Mediation

XML Wrapper XML WrapperXML Wrapper

Mediator

CM Integrated View

User/Client

Integrated View Definition IVD(S_1,S_2,…,S_k)

S_2 S_kS_1

CM Plug-ins

CM Wrapper CM Wrapper CM Wrapper

CM S_1 CM S_2 CM S_k

GCM GCM GCM

CM Queries & Results

Model-based Mediation

• “Lift” from syntax level to conceptual level• Lift:

– before: the source has element names that are NOT related

– after: the element names are linked to a domain map

• Data provider adds links from raw data to domain maps

Model-based Mediation

• CM plug-in

To make the mediator independent of CM formalism:

--Sources export all CM information in XML

--Use GCM so that the mediator no longer needs one module per CM formalism

Model-based Mediation

• CM to GCM

GCM is a meta-model that any conceivable CM formalism can be expressed in.

• F-Logic as GCM

--Convenience: root in knowledge representation and Object-Oriented database

--Availability: FLORA, FLORID

A Question

• Different data sources contains different aspects of data. How to integrate them?

For example

Calcium channel

Cell membrane

Ca++

Intracellular

Extracellular

Structural vs. Semantic Integration

• Source 1 Physiological data of calcium

current through calcium channels

• Source 2

Immunolocalization of calcium channels

•Structurally they are isolated

•Conceptually and Semantically they are related

Physiology data

Immunolocalization data

Domain Maps• Domain Map = Ontology

– definition of “things” that are relevant to your application

– representation of terminological knowledge– explicit specification of a conceptualization– concept hierarchy (“is-a”)– further semantic relationships between concepts abstractions of relational schemas, (E)ER, UML

classes, XML Schemas• Formalisms: Semantic nets, Frame-logic, Description logic, ...

Domain Maps• Formal definition

--A finite set containing: --Description Logic (DL) --Logic rules --Facts expressed as edge-labeled digraphs

with nodes representing concepts and edge labeled as roles:

C r D : if c belongs to C then there is some d in D such that r(c,d) holds

Domain Map

• Use in Model-Based Mediation --“Provide declarative means for specifying

additional knowledge that is not present in the source but that can be used to navigate through and interrelate the multiple data sources.”

--when used as part of the IVD, can infer knowledge or derive virtual relations

Brain

Neuron

Cerebellum

Purkinje cell layer

Purkinje cell

has_a

has_a

has_a

is_a

Knowledge based mediation (Use of Domain Maps)

Using ontology maps to encode these semantic relationships

Domain Maps

The Whole Picture

XML Wrapper XML WrapperXML Wrapper

Mediator

CM Integrated View

User/Client

Integrated View Definition IVD(S_1,S_2,…,S_k)

S_2 S_kS_1

CM Plug-ins

CM Wrapper CM Wrapper CM Wrapper

CM S_1 CM S_2 CM S_k

GCM GCM GCM

CM Queries & Results

Domain Map

XML-Based vs. Model-Based Mediation

Raw DataRaw DataRaw Data

IF THEN IF THEN IF THEN

LogicalDomainConstraints

Integrated-CM :=

CM-QL(Src1-CM,...)

Integrated-CM :=

CM-QL(Src1-CM,...)

. . .... (XML)

Objects

Conceptual Models

XMLElements

XML Models

C2 C3

C1

R

Classes,Relations,is-a, has-a, ...

Domain

Maps

Domain

MapsIntegrated-DTD :=

XML-QL(Src1-DTD,...)

Integrated-DTD :=

XML-QL(Src1-DTD,...)

No DomainConstraints

A = (B*|C),DB = ...

Structural Constraints (DTDs),Parent, Child, Sibling, ...

CM ~ {Descr.Logic, ER, UML, RDF/XML(-Schema), …} CM-QL ~ {F-Logic, …}

Achieving Interoperability of Genome Databases

Through Intelligent Web Mediators

Problem: There are hundreds or even thousands of biology databases, each with its own interface. Querying these databases are tedious, expensive and error prone.

Solution: Developing a database-independent, intelligent user interface using their existing query systems and architecture.

Abstraction Hierarchy of the Genome Database on the Web

accept into clustalx (select clean(a.sequence) from GlobalDB as g, AnimalDB as a where g.organism = “Drosophila” and g.source(country)=“Kenya” and g.journal like “USA” and a.accession in (select b.accession from blast(AnimalDB, clean(g.sequence)) as b where b.e-value >= 0.98))

GlobalDB

GenBankAnimalDB PlantDB

AceDB FlyBase MaizeDB RiceDB

GQL Example

LifeDB Web Browser

WebInterface

Response

WebInterface XML Negotiator

Query Processor

Interpreter

Generalizer

Mediator

Database

SchemaMappings

GlobalSchema

QueryMappings

GQL Query G Schema Query S or Query Mappings

Answer A

Web Server

Ontology

WebInterface

Response

Query G

Parameterized queries and

responses

Data queries and

responses

Query map info

map infoGlobal scheme

Query plan

GQL query

schema info

feedback loop

schema queries test data test data

probes

••••

more databases

ISIS Mediation Architecture

• ISIS : Interoperable Spatial Information System– Integration of Heterogeneous Spatial or Geographic information system.– Multi-Agent Paradigm Sharing spatial knowledge and Services.– Web Oriented Information System

– Example of Geographic information systems (GIS’s):

• Road, Traffic Information on an area• Land use information• Population Distribution• Marketing research Demographics

University of Bourgogne (France)

S2

WrapperAgent

CA

Cooperation Bus

Query Processing

Agent

InterfaceAgent

ISIS Mediation ArchitectureMULTI-agent System Architecture

OntologyAgent

S1

WrapperAgent

CA

Semantic routerAgent

USER

CA = Cooperation agent

CA

WRAPPER AGENT:

~ processes OQL (Object Query Language) queries from Corresponding

Cooperation AgentDifference between SQL, OQL: refer to this suggested

website:

http://www-db.stanford.edu/~ullman/fcdb/spr99/lec15.pdf

~ Forwards the results to the Cooperation Agent

~ A wrapper Agent is an “Employee” of one Cooperation Agent.

Responsive when triggered by the “boss”

~ Schemas are represented by AMUN (Multi-level data Model) objects

Which Lacks Semantics

COOPERATION AGENT:~ Contains knowledge of one source only ( represented by Semantic

Cooperation Objects)

~ Semantic Objects are created with the help of the semantic router agent

~ Process self initiated Queries or sub Queries initiated by other agents

~ Queries are written in terms of the local objects and passed to the wrapper

ONTOLOGY AGENT:

~ provides Mutual understanding of concepts between the various agents to

help them work with each other without a need for a global schema

~ defines ontological set of terms to be used by the cooperation agents and

the semantic router

SEMANTIC ROUTER AGENT:~ To achieve communication between Cooperation agents, The semantic

Router provides information about the location and identity of every

Cooperation agent. Cooperation agents can participate in

executing queries

Query PROCESSOR AGENT:

~ It identifies relevant information sources and creates an execution plan

INTERFACE AGENT:~ Receives Queries from the user and pass them to one Cooperation agent.

~ reports back the results of the query to the user

~ only connected with one Cooperation agent

AMUN DATA MODEL

• used to represent schemas on both the wrapper level and the cooperation level.

Geometry

Coordinate Geometry

Curve Point Surface Solid

Line String Polygon

Line ringLineISIS Page 6

Polyhedral surface

Type hierarchy of AMUN

ISIS vs Model-Based Mediation With Domain Maps

• ISIS:• Application: developed to integrate

heterogeneous geographic systems

in the first place

• Terminological Knowledge: Uses Ontology Agent

• Schemas: Represented by AMUN Data Model in all stages of mediation

• Model-Based:• Application: developed to integrate

heterogeneous Biological data bases in the first place

• Terminological Knowledge : Uses domain maps

• Schemas: represented in different models in different stages (XML,CM,GCM)

QUESTONS?COMMENTS?

References• [1] Model –based Mediation with Domain Maps, B. Ludäscher, A. Gupta, M. E. Martone, 17th

Intl. Conference on Data Engineering, Heidelberg, Germany, IEEE Computer Society, April 2001. http://www.sdsc.edu/~ludaesch/Paper/icde01.pdf

• [2].Model-Based Information Integration in a Neuroscience Mediator System, B. Ludäscher, A. Gupta, M. E. Martone, demonstration track, 26th Intl. Conference on Very Large Databases (VLDB), Cairo, Egypt, September 2000. http://www.sdsc.edu/~ludaesch/Paper/ssdbm00.html

• [3] ISIS: A Semantic Mediation Model and an Agent Based Architecture for GIS Interoperability, Eric Leclercq, Djamal Benslimane and Kokou Yétongnon, In Proceedings of the 1999 International Database Engineering and Applications Symposium, IDEAS 1999, 2 - 4 August, 1999, Montreal, Canada.

• [4]. Model-Based Mediation: Framework and Challenges, B. Ludäscher, Faculty Research Seminar, Computer Science and Engineering, U.C. San Diego, November 28th, 2001.http://www.sdsc.edu/~ludaesch/Paper/mbm-research-11-2001.ppt

• [5] Achieving interoperability of genome databases through intelligent web mediator, H. M. Jamil, In Proceedings of the IEEE International Symposium on Bio-Informatics and Biomedical Engineering (BIBE 2000), Washington, DC, November 8-10, 2000.http://www.cs.msstate.edu/~jamil/my-pub-papers/final-bibe.ps